I have problem by definition of strain and stress. From Gockenbach's book that our reference for FEM, we have $$\epsilon=\frac{\nabla u+ \nabla u^T}{2},$$
that $u$ is vector displacement, and $\nabla u$ is the Jacobian of $u$. So we have $\epsilon$ is symmetric and also $\sigma$, that is
$$2\mu \epsilon+\lambda tr(\epsilon)I$$
My problem is that I see everywhere this statement: if $\epsilon$ is symmetric or if $\sigma$ is symmetric we have... why? I can not see the case that they not be symmetric,